Hernandezine, a novel AMPK activator induces autophagic cell death in drug-resistant cancers

Drug resistance hinder most cancer chemotherapies and leads to disease recurrence and poor survival of patients. Resistance of cancer cells towards apoptosis is the major cause of these symptomatic behaviours. Here, we showed that isoquinoline alkaloids, including liensinine, isoliensinine, dauricine, cepharanthine and hernandezine, putatively induce cytotoxicity against a repertoire of cancer cell lines (HeLa, A549, MCF-7, PC3, HepG2, Hep3B and H1299). Proven by the use of apoptosis-resistant cellular models and autophagic assays, such isoquinoline alkaloid-induced cytotoxic effect involves energy- and autophagy-related gene 7 (Atg7)-dependent autophagy that resulted from direct activation of AMP activated protein kinase (AMPK). Hernandezine possess the highest efficacy in provoking such cell death when compared with other examined compounds. We confirmed that isoquinoline alkaloid is structurally varied from the existing direct AMPK activators. In conclusion, isoquinoline alkaloid is a new class of compound that induce autophagic cell death in drug-resistant fibroblasts or cancers by exhibiting its direct activation on AMPK.


INTRODUCTION
Autophagy is a highly coordinated process respon sible for maintaining normal cellular homeostasis under nutrient deprivation conditions. This process involves the lysosomal degradation of cellular components such as misfolded proteins or damaged organelles. Defects in autophagy are correlated to the pathogenesis of diseases such as cancer, myopathy and neurodegeneration [1]. AMP activated protein kinase (AMPK), maintains normal energy balance by regulating cellular metabolisms in an AMP/ADP ratiodependent manner, is responsible for the proper mechanistic modulation of autophagy [2]. During cellular starvation, AMPK induces autophagy by phosphorylating Ulk1, the mammalian counterpart of ATG1, at Ser 317 and 777 [3,4]. Molecular studies demonstrated that Ulk1 together with another mammalian ATG1 homolog, Ulk2, form complex with mATG13 and FIP200 (mammalian homologues of ATG13 and ATG17) and regulate the autophagic machinery [5,6]. Yeast models have also suggested the inductive role of ATG1 kinase in autophagy [7]. Under nutrientrich conditions, the activation of mTOR prevents the phosphorylation of Ulk1 activation through Ser 757, which finally inhibits the Ulk1AMPK dependent induction of autophagy [4]. AMPK stimulates autophagy through the inhibition of mTORC1, which is the key regulator of growth factor and nutrient signals transduction [4]. Recently, small-molecule AMPK activators have been identified as potential therapeutic agent for metabolic diseases or cancers [2,8,9]. Natural compounds such as α-Lipoic acid, polyphenols (resveratrol) and isoquinoline alkaloid (berberine); small molecule activators such as A769662, metformin, thiazolidinediones (TZDs) and alkyl benzo quinones could directly or indirectly activate AMPK in a variety of cell types [10,11]. In fact, autophagy may play its anticancer role by preventing accumulation of damaged proteins and organelles which lead to the progression of tumor growth [12], or via the induction of autophagic cell death [13].
Hernandezine, an alkaloid isolated from Chinese medicinal herb, has long been used for treating hypertension and angina pectoris [14,15]. There was report suggesting hernandezine blocks the influx of calcium via non selective cation channels in HL-60 cells [16]. Further study showed that the calcium influx triggered by depletion of internal calcium stores was blocked by hernandezine [17]. In the present study, we depicted the role of hernandezine in inducing autophagy and autophagic cell death in apoptosisresistant cells via the direct activation of AMPK.

Hernandezine exhibits specific cytotoxicity towards cancer cells
We previously demonstrated that a group of alkaloid compounds including liensinine, isoliensinine, dauricine and cepharanthine exhibit potent anticancer effect via autophagy induction [13]. Hernandezine, an alkaloid isolated from Thalictrum glandulosissimum sharing structural similarity with isoquinoline alkaloids ( Figure 1A), may also possess potent anti-cancer efficacy.

Hernandezine induces autophagic GFP-LC3 puncta in various types of cancer cells
To confirm whether hernandezine is capable of inducing autophagy in variety of cancer cells, we utilized HeLa, MCF-7, PC-3, Hep3B, A549 and H1299, and LO2 normal human hepatocytes for detecting the autophagic GFP-LC3 puncta. As shown in Figure 2A, 10 μM of hernandezine induced GFP-LC3 puncta formation in all the cancer cells and normal hepatocytes, indicating the autophagic effect of hernandezine is not cell-type specific. However, quantitation of the percentages of cells with autophagic puncta formation showed that, different cancer cell types possess different potency for autophagy induction in response to hernandezine treatment ( Figure 2B). In addition, the formation of LC3-II puncta was further verified by immunofluorescence staining against endogenous LC3-II in HeLa cancer cells ( Figure 2C). Besides, the hernandezineinduced autophagic effect was further validated with 3methyladenine (3MA), a wellknown PI3K inhibitor commonly used to inhibit autophagy [18]. As demonstrated by the decreased percentage of cells with GFP-LC3 puncta formation ( Figure 2D), addition of 3-MA abrogated hernandezineinduced autophagy.

Hernandezine induces autophagic flux in HeLa cancer cells
Induction of autophagy indicated by an increased formation of GFP-LC3 puncta using fluorescence microscopy, or LC3 lipidation using western blot, can be resulted from either an induction of autophagic flux or failure in fusion of autophagosomes and lysosomes. Hence, we measured the conversion of soluble LC3-I to lipid-bound LC3-II in the presence of E64d and pepstatin A, which inhibit lysosomal proteases including cathepsins B, D and L; or bafilomycin, which inhibits the fusion of autophagosome and lysosome by raising lysosomal pH [19,20]. As expected, hernandezine increased the rate of LC3-II formation in the presence of the inhibitors when compared with the use of inhibitors or hernandezine alone ( Figure 3A and 3B). This result suggested that hernandezine induced autophagic activity through enhanced autophagic flux and autophagosome formation.
We further monitored the autophagic flux using mRFP-GFP tandem fluorescent-tagged LC3 (tfLC3) plasmid. Given that the localisation pattern of GFP-LC3 presence of lysosomal inhibitors. HeLa cells were treated with 10 µM of hernandezine in the presence or absence of 10 mg/mL lysosomal protease inhibitors (E64d and pep. A) for 24 h, or 50 nM bafilomycin A for 8 h. Cell lysates were analysed by western blot for LC3 conversion (LC3-I, 18 kDa; LC3-II, 16 kDa) and β-actin. LC3-II band intensities were quantified using densitometric analysis and normalised to β-actin. Data were expressed as a fold change relative to the DMSOtreated negative control. Bar charts were representatives of three independent experiments. (C) mRFP-GFP-LC3 fluorescence localisation pattern of hernandezine. HeLa cells were firstly transfected with the mRFP-GFP-LC3 plasmids for 24 h and then treated with 10 μM of hernandezine for 0-24 h. Cells were then subjected to immunocytochemical analysis and the mRFP + -GFP + (yellow) puncta were scored; scale bar, 15 mm. Each correlation plot was derived from the field shown in the immunofluorescence image. The colocalisation of mRFP with GFP signal from tfLC3 puncta was measured using ImageJ software. The percentage of colocalisation was displayed in the bar chart. The values indicated the average of at least five images. Error bars, S.D., ***P < 0.001. and tfLC3 are different, the LC3 fusion construct with both red (mRFP) and green (GFP) fluorescence proteins is therefore widely used for detection of autophagosomes [21]. Due to the difference in the stability of GFP and mRFP under different pH conditions [22], acidic environment of lysosome will quench the GFP signal but not the mRFP signal. Therefore, we overexpressed the tfLC3 plasmid to monitor autophagic flux. As shown in Figure 3C, while the yellow merged image (mRFP+GFP+) represents the autophagosomes, merged images with red puncta (mRFP+-GFP−) indicates autophagic flux with the formation of autolysosomes [21]. Our results demonstrated a timedependent decrease in the percentage of cells with mRFPGFP colocalisation after hernandezine treatment, confirming the induction of autophagic flux by this alkaloid in HeLa cancer cells.

Hernandezine activates AMPK kinase for induction of autophagy and cell death
AMPK is a sensor of cellular energy status and is activated under high intracellular AMP conditions such as hypoxia or nutrient deprivation, thereby induces autophagy via the mTORdependent pathway [23]. Phosphorylation of AMPK and its downstream target AcetylCoenzyme A Carboxylase (ACC) are required for smallmolecule induced autophagy [24]. As demonstrated by western blot analysis, AMPK phosphorylation was increased in response to hernandezine treatments ( Figure 4A). The phosphorylation was accompanied by a reduction in phosphorylated p70S6K, a downstream target of mTOR ( Figure 4A). Concomitantly, ACC, the direct downstream target of AMPK, was phosphorylated upon hernandezine treatments ( Figure 4A, lower panel). In addition, a decrease in hernandezine-induced GFP-LC3 autophagic puncta formation was observed in cells pretreated with the AMPK inhibitor compound C (CC) ( Figure 4B), suggesting the involvement of AMPK signalling in hernandezineinduced autophagy. Alternatively, supplementation of glycolytic intermediate, methyl pyruvate (MP), was able to suppress hernandezine induced LC3-II conversion and GFP-LC3 puncta formation ( Figure 4C and 4D), suggesting hernandezine induced autophagy involved energy depletion. To address whether hernandezineinduced cell death is due to energy depletion, we examined its cytotoxicity with the presence of methyl pyruvate using annexin V flow cytometry. As shown in Figure 4E, while hernandezine induced cell death in HeLa cancer cells, the addition of methyl pyruvate abolished the compoundinduced cell death. Most importantly, cellfree AMPK kinase assay has revealed that liensinine, isoliensinine, dauricine, cepharanthine and hernandezine, could activate the AMPK kinase activity dosedependently ( Figure 4F). All these evidence suggested that isoquinoline alkaloid activates on AMPK kinase directly.
To determine whether hernandezine requires Atg7 for autophagy induction, GFP-LC3 transfected Atg7-wild-type and -deficient mouse embryonic fibroblasts (MEFs) were incubated with hernandezine for 24 h. The hernandezine treated MEFs were then fixed for quantification of GFP-LC3 puncta formation. As shown in Figure 5A, 10 μM of hernandezine increased GFP-LC3 puncta formation in Atg7 wild-type MEFs, but not in Atg7-deficient MEFs, indicating the involvement of Atg7 in hernandezine induced activation of autophagy.
A number of anticancer agents have been reported to induce autophagy in various types of cancers [27], however it remains controversial whether autophagy promotes cell death or acts as a prosurvival mechanism. Studies showed that Atg7-deficient MEFs are resistant to induction of autophagy [25]. As hernandezineinduced autophagy required Atg7 ( Figure 5A), we therefore utilized both Atg7 wild-type and Atg7-deficient MEFs to determine whether hernandezineinduced autophagy leads to cell death or acts as a prosurvival mechanism [19]. Results showed that hernandezine exhibited less cytotoxicity in Atg7-deficient MEFs when compared to their wildtype counterparts ( Figure 5B), similar results were found in HeLa cancer cells with Atg7 knockdown ( Figure 5C). These data suggested that hernandezine induced autophagy could lead to autophagic cell death, because the failure in the induction of autophagy in Atg7 deficient cells suppressed the hernandezine-induced cell death. Hernandezineinduced autophagy requires the involvement of Atg7 and promotes cell death in cancer cells.

Hernandezine induces autophagic cell death in apoptosis-resistant cancer cells
Cancer cells are frequently resistant to drugmediated apoptosis [28]. Therefore, the use of smallmolecules to induce autophagic cell death in apoptosisdefective or apoptosisresistant cancer cells may become a prom ising therapeutic approach [13,29]. To investigate if the identified AMPK activator hernandezine can exhibit cytotoxic effects towards apoptosisresistant cells, we adopted a panel of apoptosisdefective or apoptosis resistant cells such as caspase 3/-7/-8 deficient MEFs, Bax-Bak double knockout (DKO) MEFs and DLD-1 Bax-Bak DKO human colon cancer cells as the cellular models. As shown in Figure 6A, hernandezine demonstrated similar cytotoxic profiles in caspase -3/-7 DKO, wild-type and caspase -3/-7/-8 deficient MEFs. Similar cytotoxic effect towards both BaxBak wildtype and DKO MEFs or DLD-1 colon cancer cells were also observed. Bax-Bak DKO MEFs revealed resistance towards chemotherapeutic agents, i.e. cisplatin, adriamycin, taxol, etoposide and staurosporine with resistance factors ranging from 2.6 to 27.6 ( Figure 6B), suggesting that hernandezine could circumvent the apoptosisresistant phenotype of cells conferred by genetic deficiencies. We then examined the cytotoxic effects of hernandezine in both BaxBak wild type and DKO MEFs using annexin V flow cytometry analysis. As expected, there was coherence between the MTT and flow cytometry results, which suggested that hernandezine induces potent cytotoxicity in apoptosis defective or apoptosisresistant cells ( Figure 6C). Owing to the direct activation of AMPK by hernandezine, we also determined the role of AMPK in hernandezineinduced autophagic cell death in BaxBak DKO apoptosisresistant cells. Consistently, AMPK inhibitor compound C (CC) suppressed the hernandezineinduced autophagy and cell death in BaxBak DKO MEFs ( Figures 6D and 7A), whereas CC also abrogated hernandezineinduced cell death in DLD-1 Bax-Bak DKO cancer cells ( Figure 7B), confirming the key role of AMPK signalling in hernandezineinduced cell death of apoptosisresistant cancer. Furthermore, the multidrugresistant (MDR) cancer cells were also adopted to evaluate the potential anticancer effect of hernandezine. For this purpose, taxol resistant HCT8 colon cancers were incubated with 10 µM of hernandezine in the presence of CC prior to annexin V flow cytometry analysis. Addition of CC blocked the hernandezineinduced cell death in these MDR cancer cells ( Figure 7C).

DISCUSSION
Natural alkaloids namely isoquinoline alkaloids comprising the common structure of isoquinoline nucleus, have been shown to possess anticancer properties as demonstrated by their cytotoxic effect on various types of cancer cells [13]. These alkaloids such as liensinine, isoliensinine, dauricine and cepharanthine, trigger cell death in a nonapoptotic manner, therefore, immortalized cell lines with hampered apoptosis are sensitized to their stimulation. This alkaloidinduced cellular toxicity is associated with the upregulation of Atg7dependent autophagy, which is potentially beneficial to anti-cancer therapy [13]. Although the molecular mechanisms underpinning the alkaloid induced cell death is still elusive, we and others have demonstrated that the AMPKmTOR signaling cascade is activated by these alkaloids [13,[30][31][32].
Although other natural compounds such as coenzyme Q (CoQ), and polyphenol including flavonoids, lignans, stilbenes and phenolic acids were found intertwining with the AMPK signaling pathway [33][34][35], there is lack of evidence pointing towards the direct activation of AMPK by these compounds. CoQ and polyphenol activate AMPK signaling via the upstream kinases, Ca 2+ -stimulated kinase (CaMKK) and liver kinase B1 (LKB1), which are correlated to their therapeutic potency upon different disease conditions such as obesity, hyperglycemia and insulin resistance [33][34][35]. Other studies showed that genistein (flavonoid) contributes to obesity control by regulating the transcriptional expression of fatty acid ω-hydroxylase (CYP4F2) through the manipulation of CaMKK [36]. Sauchinone (lignan) activates AMPK phosphorylation by LKB1 kinase, which perturbed the ironinduced oxidative mitochondrial stress and lead to the alleviation of chronic disorders progression [37]. Also, application of CoQ 10 to the 3T3-L1 adipocytes culture have demonstrated the antiadipogenic effect of CoQ is mechanistically relevant to the CaMKKβ-AMPKaxisdriven peroxisome proliferatoractivated receptor alpha (PPARα) expression [38].
AMPK activators are ideal pharmacological com pounds for cancer therapy, since mTOR kinase that frequently activated in a wide spectrum of tumors are negatively regulated by the LKB1-AMPK pathway [39][40][41]. Both in vitro and animal studies have highlighted the anticancer effect of direct activation of AMPK. MT 63-78 (Debio 0903), a direct AMPK activator, thwarts the growth of androgen sensitive and castration resistant prostate cancer cell model (CRPC), and reduces tumor volume of mice intraperitoneally (i.p.) or orally treated with the compound [42]. Recent studies showed that BL-AD008 is a novel dualtarget activator of AMPK/ZIPK and induces apoptosis in cervical cancer [43]. Commonly prescribed antidiabetic drug, metformin, functioning through direct AMPK activation has epidemiologically been proven to downscale the occurrence of pancreas, colon and hepatocellular carcinoma in type 2 diabetic patients [44][45][46][47]. Other studies demonstrated that metformin potentiates anticancer effect of dasatinib in head and neck squamous cell carcinoma cells via AMPK dependent ER stress [48]. It is noteworthy that, metformin has already been approved in clinical trials for the treatment of pancreatic and breast cancers (http://clinicaltrials.gov, IDs NCT01210911, NCT01266486). The association of high glucose uptake with elevated cancer cell proliferation also reinforced the notion of applying AMPK activators clinically for cancer therapy and prevention [49]. Higher mortality rate and cancer risk are linked to the pathological condition of excessive circulatory glucose concentration such as hyperglycemia [50]. AMPK activation in hernandezinetreated cancer cells was induced when the cellular energy state is suppressed, suggesting the close relationship between metabolic glucose anomalies and the pharmacological action of hernandezine towards cancer cell death. However, activation of AMPK may protect cancer cells in response to the microenvironmental stresses that the cancer cells are encountering [51,52]. Therefore, clinical trials designated to define the optimal clinical stages for AMPK activator application is needed for maximizing their efficacy.
In line with our previous findings that isoquinoline alkaloid is able to induce autophagic cell death in cancer cells [13], hernandezineinduced cytotoxicity is autophagydependent. Similar to other isoquinoline alkaloids, hernandezineinduced cytotoxicity is independent of apoptosis. By using a wide spectrum of caspase (-3/-7/-8) and Bax/Bak-deficient cell lines, we concluded that hernandezine may activate autophagic cell death without BaxBak or caspases. Apoptotic related mitochondrial/cytochrome c pathway is frequently disrupted in human cancers and many malignancies [27,53]. Most chemotherapyresistant cancers are having defective apoptotic pathways. For example, the BaxBak doubleknockout MEFs are resistant to various apoptosis inducing agents [53]. Caspase3, 8 and 9 are associated with apoptosisresistance and drugresistance phenotypes, as well as apoptosis induced by anticancer agents [54]. Although caspase3 activation is crucial to apoptosis, study showed that the induction of apoptosis could be happened in the absence of caspase3 [55]. Recent studies further highlighted that induction of autophagy for treatment of cisplatinresistant and p53 mutated cancers [56]. Therefore, novel pharmaceutical interventions inducing cancer cells cytotoxicity through nonapoptotic signaling is inevitable. Hernandezine or generally isoquinoline alkaloids may actually serve more than simply an anti cancer agent. AMPK is engaged with glucose and lipid metabolisms extensively in different organs and tissues, controlling pancreatic insulin secretion, fatty acid and cholesterol synthesis in liver, lipolysis in adipose tissue, cardiac and skeletal fatty acid oxidation, and glucose uptake [57][58][59][60]. Apart from glucose concentration, factors like hormones and cytokines are stimuli of AMPK, illustrating the involvement of multiple pathways in AMPK regulation [61][62][63]. Therefore, AMPK is positioned at the center of AMPKmTOR cascade making the kinase the key molecular target for pharmaceutical interventions of different metabolic disorders.
Literatures concerning the direct action of natural compounds on AMPK are scarce. Up to 2012, around 26 patent applications claiming the discovery of direct AMPK activators have been filed [64]. These smallmolecules AMPK activators belong to the derivatives of thienopyridone, cyclic benzimidazole, pyrimidine, alkene oxindole and ringfused imidazole [64]. Accurate examination disclosing that they are reminiscent to each other due to the 4(2hydroxypheny)phenylside chain and a negatively ionizable group. These structural resemblances seem not to be a mandatory criteria for AMPK activation, because structural modifications of this particular side chain do not induce notable differences in AMPK activation [64]. However, the robust AMPK activation induced by hernandezine, liensinine, isoliensinine, dauricine and cepharanthine agreed with this observation as they do not contain the 4(2hydroxypheny) phenylside chain. Accordingly, we have proposed a new class of compound exhibiting direct activation of AMPK kinase, and widen the chemical scope for searching new direct AMPK activators.
Isoquinoline alkaloid is a direct activator of AMPK and autophagy, and exhibits its anticancer property by inducing cancer cell death. Provided that our candidates are structurally different from other proprietary direct AMPK activators, the present study unveils a novel class of natural smallmolecule directly activating AMPK which induces autophagy particularly on apoptoticresistance cancer.

Cell culture
All cells were obtained from the American Type Culture Collection (Rockville, MD) unless otherwise specified. Immortalised wild type and Atg7-deficient mouse embryonic fibroblasts (MEF) were kindly provided by Prof. Masaaki Komatsu (Juntendo University, School of Medicine, Japan). Immortalised wild-type and Caspase 3/7-deficient MEFs were gifts from Prof. Richard A. Flavell (Yale University School of Medicine, United States). Immortalised wild type and Caspase 8-deficient MEFs were kindly provided by Prof. Kazuhiro Sakamaki (Kyoto University, Graduate School of Biostudies, Japan). Immortalised wildtype and BaxBak double knockout MEFs were kindly provided by Prof. Shigeomi Shimizu (Tokyo Medical and Dental University, Medical Research Institute, Japan). Human DLD-1 Bax-Bak wild-type and deficient isogenic colon cancer cells were purchased from Sigma (MO, USA). HCT8 taxolresistant cancer cells were purchased from KeyGEN BioTECH (Shanghai, China). All cells were cultured with medium supplemented with 10% foetal bovine serum (FBS), 50 U/ml penicillin, and 50 mg/ml streptomycin (Invitrogen, Paisley, Scotland, UK). All cell cultures were incubated in a humidified incubator at 37ºC with 5% CO 2 .

Quantification of GFP-LC3 puncta formation
GFP-LC3 puncta were quantified as described previously [26]. Localisation of GFP-LC3 and the fluorescent images were acquired using high magnification widefield epifluorescence microscopy. Images were captured by a Photometrics CoolSNAP HQ2 CCD camera on the Olympus IX71Applied Precision DeltaVision restoration microscope (Applied Precision, Inc, USA), and the epifluorescence images were numerically deconvolved using DeltaVision algorithms (Applied Precision, Inc.). To quantify autophagy, the percentage of autophagic cells was calculated by counting the number of cells showing increased punctate pattern of GFP-LC3 and dividing by the total number of GFPpositive cells. A minimum of 1000 cells from randomly selected fields was scored per condition per experiment.

Endogenous autophagy detection
Hernandezinetreated cancer cells on cover slips were fixed with 4% paraformaldehyde (Sigma) and then immersed in methanol for 2 min. Cells were then incubated with anti-LC3 antibody (1:200) in TBST (100 mM Tris HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20 and 5% BSA) overnight at 4°C. Cells were incubated with antimouse secondary antibody (TRITC) (1:200) in TBST containing 5% BSA at 37°C for 1 h in the dark. The coverslips were mounted with FluorSave ™ mounting media (Calbiochem, San Diego, CA, USA) for fluorescence imaging. Localization of LC3 autophagosomes were captured under the API Delta Vision Live-cell Imaging System (Applied Precision Inc., GE Healthcare Company, Washington, USA). Standard guidelines were followed to monitor autophagy [20]. The percentage of autophagic cells was calculated by counting the number of the cells showing increased punctuate pattern of LC3 fluorescence (≥ 10 dots/cell) in immunofluorescence positive cells over the total number of cells in the same field. A minimum of 1000 cells from randomly selected fields were scored.

mRFP-GFP tandem fluorescent-tagged LC3 (tfLC3) immunocytochemistry and fluorescence microscopy
HeLa cells were transfected with mRFP-GFP-LC3 for 24 h. After transfection, the cells were treated with hernandezine at 10 µM for 0-24 h. Each correlation plot was derived from the field shown in the fluorescence microscopic image. Colocalization of mRFP with GFP in tfLC3 puncta was measured using ImageJ software, and presented as the percentage of the total number of yellow mRFP+-GFP+ puncta [21].

MTT cytotoxicity assays
Hernandezine was dissolved in DMSO to a final concentration of 100 mM. Cell viability was measured using the MTT (3[4, 5dimethylthiazol2yl]2, 5 diphenyl tetrazolium bromide) assay as described previously [65]. The percentage of viable cells was calculated using the following formula: Cell viability (%) = Cells number treated / Cells number DMSO control × 100. Data were obtained from three independent experiments.

Flow cytometry analysis
Cell viability and cell death were measured using an annexin V staining kit (BD Biosciences, San Jose, CA, USA). Briefly, cells were treated with 10 μM of hernandezine for 24 h. Cells were then analysed by multiparametric flow cytometry using FITC-Annexin V and Propidium iodide staining (BD Biosciences, San Jose, CA, USA). Flow cytometry was then carried out using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA, USA). Data acquisition and analysis was performed with CellQuest (BD Biosciences, San Jose, CA, USA). Data were obtained from three independent experiments.

Western blot analysis
Cells were treated with 10 µM of hernandezine for 24 h at 37ºC. After SDS/PAGE electrophoresis, the proteins from SDS/PAGE were electrotransferred to a membrane. The membrane was then immunoblotted with the appropriate antibodies. followed by HRP-conjugated secondary antibody for 60 min. Band intensities were quantified with ImageJ (N.I.H.).

AMPK kinase assay
AMPK kinase assay was performed using CycLex ® AMPK Kinase Assay Kit (MBL, Japan) according to manufacturing instructions. In brief, 0.2 ng of AMPK (α1/β1γ1) active enzyme (CycLex Co., Ltd.) was incubated in well with 10X of hernandezine (50 & 100 μM) or 10X of positive control, AMP (100 µM) in kinase assay buffer (50 μM ATP & 10 mM DTT) at 30 o C for 20 min. The reaction was then stopped by washing with buffer for 5 times. Then, antiphosphomouse IRS1 S789 monoclonal antibody was added to each well at room temperature for 30 min. After washing with buffer for 5 times, HRP conjugated anti-mouse IgG was added to each well at room temperature for 30 min. After washing with wash buffer, the TMB substrate reagent was incubated in wells at room temperature for 5-15 min. Stop solution was added to each well before measuring absorbance at 450/550 nm.

Statistical analysis
The results were expressed as the means ± SD as indicated. Differences were considered statistically significant when the Pvalue was less than 0.05. Student's ttest or oneway ANOVA analysis was used for comparison among different groups.

CONFLICTS OF INTEREST
The authors declare no conflicting financial interests.